U.S. patent number 10,578,446 [Application Number 15/121,576] was granted by the patent office on 2020-03-03 for route search apparatus, route search method and computer-readable storage medium storing program.
This patent grant is currently assigned to AISIN AW CO., LTD., TOYOTA JIDOSHA KABUSHIKI KAISHA, ZENRIN CO., LTD.. The grantee listed for this patent is AISIN AW CO., LTD., TOYOTA JIDOSHA KABUSHIKI KAISHA, ZENRIN CO., LTD.. Invention is credited to Tomoko Arita, Atsushi Ikeno, Yoshitaka Kato, Tomoki Kodan, Sadahiro Koshiba, Kazuteru Maekawa, Tomohiko Masutani, Motohiro Nakamura, Hiroyuki Tashiro, Koichi Ushida.
![](/patent/grant/10578446/US10578446-20200303-D00000.png)
![](/patent/grant/10578446/US10578446-20200303-D00001.png)
![](/patent/grant/10578446/US10578446-20200303-D00002.png)
![](/patent/grant/10578446/US10578446-20200303-D00003.png)
![](/patent/grant/10578446/US10578446-20200303-D00004.png)
![](/patent/grant/10578446/US10578446-20200303-D00005.png)
![](/patent/grant/10578446/US10578446-20200303-D00006.png)
![](/patent/grant/10578446/US10578446-20200303-D00007.png)
![](/patent/grant/10578446/US10578446-20200303-D00008.png)
![](/patent/grant/10578446/US10578446-20200303-D00009.png)
![](/patent/grant/10578446/US10578446-20200303-D00010.png)
View All Diagrams
United States Patent |
10,578,446 |
Masutani , et al. |
March 3, 2020 |
Route search apparatus, route search method and computer-readable
storage medium storing program
Abstract
A route search apparatus configured to search a route from a set
place of departure to a set destination comprises network data that
includes nodes and links representing a road network, an average
cost value indicating an average of travel time of each of the
links, and a variance value indicating a degree of variance of the
travel time, and a route searcher configured to determine the route
from the place of departure to the destination as a recommended
route, based on the average cost value, the variance value and a
weight coefficient of the variance value.
Inventors: |
Masutani; Tomohiko (Kitakyushu,
JP), Tashiro; Hiroyuki (Kitakyushu, JP),
Arita; Tomoko (Kitakyushu, JP), Nakamura;
Motohiro (Okazaki, JP), Kodan; Tomoki (Nagoya,
JP), Ikeno; Atsushi (Minato, JP), Kato;
Yoshitaka (Minato, JP), Koshiba; Sadahiro
(Takahama, JP), Maekawa; Kazuteru (Miyoshi,
JP), Ushida; Koichi (Okazaki, JP) |
Applicant: |
Name |
City |
State |
Country |
Type |
ZENRIN CO., LTD.
TOYOTA JIDOSHA KABUSHIKI KAISHA
AISIN AW CO., LTD. |
Kitakyushu-shi
Toyota-shi
Anjo-shi |
N/A
N/A
N/A |
JP
JP
JP |
|
|
Assignee: |
ZENRIN CO., LTD.
(Kitakyushu-shi, JP)
TOYOTA JIDOSHA KABUSHIKI KAISHA (Toyota-shi, JP)
AISIN AW CO., LTD. (Anjo-shi, JP)
|
Family
ID: |
54008551 |
Appl.
No.: |
15/121,576 |
Filed: |
February 19, 2015 |
PCT
Filed: |
February 19, 2015 |
PCT No.: |
PCT/JP2015/000798 |
371(c)(1),(2),(4) Date: |
August 25, 2016 |
PCT
Pub. No.: |
WO2015/129214 |
PCT
Pub. Date: |
September 03, 2015 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20160363455 A1 |
Dec 15, 2016 |
|
Foreign Application Priority Data
|
|
|
|
|
Feb 27, 2014 [JP] |
|
|
2014-036275 |
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G01C
21/3453 (20130101); G08G 1/096827 (20130101); G08G
1/096838 (20130101) |
Current International
Class: |
G01C
21/34 (20060101); G08G 1/0968 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
1 614 996 |
|
Jan 2006 |
|
EP |
|
2005-91303 |
|
Apr 2005 |
|
JP |
|
2008-241605 |
|
Oct 2008 |
|
JP |
|
2012-141145 |
|
Jul 2012 |
|
JP |
|
Other References
International Search Report dated May 19, 2015 in PCT/JP2015/000798
filed Feb. 19, 2015. cited by applicant.
|
Primary Examiner: Mancho; Ronnie M
Attorney, Agent or Firm: Oblon, McClelland, Maier &
Neustadt, L.L.P.
Claims
The invention claimed is:
1. A route search apparatus configured to search a route from a set
place of departure to a set destination, the route search apparatus
comprising: a memory device under control of process circuitry and
configured to store network data that includes nodes and links
representing a road network, an average cost value indicating an
average of travel time of each of the links, and a variance value
indicating a degree of variance of the travel time; and the process
circuitry configured to determine a plurality of halfway route
candidate routes from the place of departure to a node
corresponding to a specific point in the middle of the route from
the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through, perform
a first determination process that identifies from the plurality of
halfway route candidate routes a first halfway route candidate
having a smallest candidate overall cost value out of the
respective candidate overall cost values, determines whether the
destination is located at a last link or last node of the first
halfway route candidate, and fix the first halfway route candidate
as the recommended route when it is determined that the destination
is located at the last link or the last node of the first halfway
route candidate.
2. The route search apparatus according to claim 1, wherein the
process circuitry calculates for the plurality of halfway route
candidates the respective overall cost values by adding a
correction value calculated as a product of the weight coefficient
and a value having a positive correlation to an integrated value of
the variance values corresponding to links that are passed through
between the place of departure and the destination, to an
integrated value of the average cost values corresponding to the
links.
3. The route search apparatus according to claim 1, wherein when
there are a plurality of the respective candidate overall cost
values that are different from each other by at most a
predetermined value, the process circuitry determines the first
halfway route candidate, based on one of the first term and the
second term that is selected according to the weight
coefficient.
4. The route search apparatus according to claim 1, wherein the
process circuitry is further configured to: perform a second
determination process when there is at least one provisional second
halfway route candidate having a smaller integrated value of the
average cost values than an integrated value of the average cost
values of the first halfway route candidate, out of remaining
halfway route candidates that are the halfway route candidates
other than the first halfway route candidate, and deter lines a
second halfway route candidate having a smallest candidate overall
cost value out of at least one provisional second halfway route
candidate; perform a third determination process that specifies the
second halfway route candidate determined by the second
determination process, as the first halfway route candidate,
specifies the halfway route candidate other than the determined
first halfway route candidate and second halfway route candidate,
as the remaining halfway route candidate, and repeats the second
determination process, and identify the first halfway route
candidate and the second halfway route candidate determined by the
first to the third determination processes as halfway routes.
5. The route search apparatus according to claim 1, wherein the
process circuitry processes statistical information indicating
histograms of the travel time of respective links corresponding to
roads that are passed through in each of the halfway route
candidates, by convolution operation, so as to generate candidate
statistical information indicating a histogram of the travel time
with regard to each of the halfway route candidates, and the
process circuitry identifies the first halfway route out of the
plurality of halfway route candidates based on the respective
candidate overall cost values calculated according to a function
including the weight coefficient and a candidate average cost value
representing an average of the travel time of each of the halfway
route candidates and a candidate variance value representing a
degree of variance of the travel time of the halfway route
candidate that are calculated from the candidate statistical
information.
6. The route search apparatus according to claim 5, wherein the
process circuitry calculates the respective candidate overall cost
values according to a function including a first term representing
the candidate average cost value and a second term representing a
correction value calculated based on the candidate variance value
and the weight coefficient.
7. The route search apparatus according to claim 1, wherein the
process circuitry determines the recommended route with regard to
each of a plurality of different values of the weight
coefficient.
8. The route search apparatus according to claim 5, wherein the
process circuitry calculates an index indicating a degree of
variance of the travel time of the recommended route, based on a
standard deviation of the generated statistical information.
9. The route search apparatus according to claim 1, wherein the
average cost value and the variance value with regard to each of
the links are calculated based on original information regarding
travel time data of the travel time and a probability of each
travel time, and when the travel time of a link is affected by a
feature at a certain frequency, the average cost value of a
specific link that is the link affected by the feature is
calculated from the entire travel time data and all the
probabilities included in the original information, and the
variance value of the specific link is calculated from the travel
time data and the probability that are estimated to be not affected
by the feature in the original information.
10. A route search method of searching a route from a place of
departure to a destination, comprising: storing network data in a
memory device under control of process circuitry, the network data
including nodes and links representing a road network, an average
cost value indicating an average of travel time of each of the
links, and a variance value indicating a degree of variance of the
travel time; determining using the processing circuitry a plurality
of halfway route candidate routes from the place of departure to a
node corresponding to a specific point in the middle of the route
from the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through,
performing a first determination process that identifies from the
plurality of halfway route candidate routes a first halfway route
candidate having a smallest candidate overall cost value out of the
respective candidate overall cost values, determining whether the
destination is located at a last link or last node of the first
halfway route candidate, and fixing the first halfway route
candidate as the recommended route when it is determined that the
destination is located at the last link or the last node of the
first halfway route candidate.
11. A non-transitory computer readable storage medium storing a
program configured to cause process circuitry of a computer to
implement a function of searching a route from a place of departure
to a destination, the program causing the computer to implement the
functions of: storing network data in a memory device under control
of the process circuitry, the network data including nodes and
links representing a road network, an average cost value indicating
an average of travel time of each of the links, and a variance
value indicating a degree of variance of the travel time;
determining using the processing circuitry a plurality of halfway
route candidate routes from the place of departure to a node
corresponding to a specific point in the middle of the route from
the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through,
performing a first determination process that identifies from the
plurality of halfway route candidate routes a first halfway route
candidate having a smallest candidate overall cost value out of the
respective candidate overall cost values, determining whether the
destination is located at a last link or last node of the first
halfway route candidate, and fixing the first halfway route
candidate as the recommended route when it is determined that the
destination is located at the last link or the last node of the
first halfway route candidate.
12. A route search apparatus configured to search a route from a
set place of departure to a set destination, the route search
apparatus comprising: a memory device under control of process
circuitry and configured to store network data that includes nodes
and links representing a road network, an average cost value
indicating an average of travel time of each of the links, and a
variance value indicating a degree of variance of the travel time;
and the process circuitry configured to determine a plurality of
halfway route candidate routes from the place of departure to a
node corresponding to a specific point in the middle of the route
from the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through, perform
a first determination process that identifies from the plurality of
halfway route candidate routes a first halfway route candidate
having a smallest candidate overall cost value out of the
respective candidate overall cost values, determines whether the
destination is located at a last link or last node of the first
halfway route candidate, fix the first halfway route candidate as
the recommended route when it is determined that the destination is
located at the last link or the last node of the first halfway
route candidate, and control display of the recommended route on a
display panel.
13. A route search method of searching a route from a place of
departure to a destination, comprising: storing network data in a
memory device under control of process circuitry, the network data
including nodes and links representing a road network, an average
cost value indicating an average of travel time of each of the
links, and a variance value indicating a degree of variance of the
travel time; determining using the processing circuitry a plurality
of halfway route candidate routes from the place of departure to a
node corresponding to a specific point in the middle of the route
from the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through,
performing a first determination process that identifies from the
plurality of halfway route candidate routes a first halfway route
candidate having a smallest candidate overall cost value out of the
respective candidate overall cost values, determining whether the
destination is located at a last link or last node of the first
halfway route candidate, fixing the first halfway route candidate
as the recommended route when it is determined that the destination
is located at the last link or the last node of the first halfway
route candidate, and controlling, using the processing circuitry,
display of the recommended route on a display panel, the display
panel being a component of a hand held device, a personal computer,
or a car navigation system.
14. A non-transitory computer readable storage medium storing a
program configured to cause process circuitry of a computer to
implement a function of searching a route from a place of departure
to a destination, the program causing the computer to implement the
functions of: storing network data in a memory device under control
of the process circuitry, the network data including nodes and
links representing a road network, an average cost value indicating
an average of travel time of each of the links, and a variance
value indicating a degree of variance of the travel time;
determining using the processing circuitry a plurality of halfway
route candidate routes from the place of departure to a node
corresponding to a specific point in the middle of the route from
the place of departure to the destination based on respective
candidate overall cost values for each of the plurality of halfway
route candidates that are calculated as a sum of a first term
representing an integrated value of the average cost values
corresponding to links that are passed through from the place of
departure to the node corresponding to the specific point in the
middle of the route from the place of departure to the destination
and a second term representing a correction value calculated based
on a weight coefficient and an integrated value of the variance
values corresponding to the links that are passed through,
performing a first determination process that identifies from the
plurality of halfway route candidate routes a first halfway route
candidate having a smallest candidate overall cost value out of the
respective candidate overall cost values, determining whether the
destination is located at a last link or last node of the first
halfway route candidate, fixing the first halfway route candidate
as the recommended route when it is determined that the destination
is located at the last link or the last node of the first halfway
route candidate, and controlling, using the processing circuitry,
display of the recommended route on a display panel, the display
panel being a component of a hand held device, a personal computer,
or a car navigation system.
15. The route search apparatus according to claim 12, wherein the
display panel is a component of a hand held device, a personal
computer, or a car navigation system.
16. The route search method according to claim 13, wherein the
display panel is a component of a hand held device, a personal
computer, or a car navigation system.
17. The non-transitory computer readable storage medium according
to claim 14, wherein the display panel is a component of a hand
held device, a personal computer, or a car navigation system.
18. The route search apparatus according to claim 1, wherein the
process circuitry is further configured to extend a search tree
from the last link or the last node of the first halfway route
candidate to the destination when it is determined that the
destination is not located at the last link or the last node of the
first halfway route candidate.
19. The route search method according to claim 10, wherein the
process circuitry is further configured to extend a search tree
from the last link or the last node of the first halfway route
candidate to the destination when it is determined that the
destination is not located at the last link or the last node of the
first halfway route candidate.
20. The non-transitory computer readable storage medium according
to claim 11, wherein the process circuitry is further configured to
extend a search tree from the last link or the last node of the
first halfway route candidate to the destination when it is
determined that the destination is not located at the last link or
the last node of the first halfway route candidate.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
The present application claims priority from Japanese patent
application P2014-36275 filed on Feb. 27, 2014, the content of
which is hereby incorporated by reference into this
application.
TECHNICAL FIELD
The present invention relates to a technique of searching a route
from a place of departure to a destination.
BACKGROUND ART
Car navigation system mounted on the automobile, cell phones,
handheld game consoles, PNDs (Personal Navigation Devices) and PDAs
(Personal Digital Assistants) have recently been known as the route
search apparatus configured to search a route from a place of
departure to a destination as described in, for example, JP
2012-141145A and JP 2008-241605A.
For example, the technique disclosed in JP 2012-141145A uses a
reference travel time and a variance value representing a variance
of the travel time that are set with regard to each link, to search
a route from a place of departure to a destination. More
specifically, this technique calculates an expected time required
for a guide route by sequential summation of the reference travel
times of the respective links constituting the guide route, and
calculates the probability of the expected time by sequential
summation of the variance values of the respective links.
SUMMARY
Technical Problem
In the technique disclosed in JP 2012-141145A, however, a fixed
value is set as the variance value with regard to each link. This
may provide a result of route search that lacks flexibility. For
example, the user may require route search by taking into account
only the reference travel time or may require route search by
taking into account the variance value. Even in the case of route
search by taking into account the variance value, there is a demand
for changing the degree of the variance value in calculation of the
result of route search. Other needs over the prior art include, for
example, improvement of the processing efficiency, downsizing of
the apparatus, cost reduction, resource saving and improvement of
the convenience.
Solution to Problem
In order to solve the problems described above, the invention may
be implemented by aspects or applications described below.
(1) According to one aspect of the invention, there is provided a
route search apparatus configured to search a route from a set
place of departure to a set destination. This route search
apparatus may comprise a storage part configured to store network
data that include nodes and links representing a road network, an
average cost value indicating an average of travel time of each of
the links, and a variance value indicating a degree of variance of
the travel time; and a route searcher configured to determine the
route from the place of departure to the destination as a
recommended route, based on an overall cost value calculated
according to a function including the average cost value, the
variance value and a weight coefficient of the variance value. The
route search apparatus of this aspect can determine the recommended
route by taking into account the weight coefficient of the variance
value, thus allowing for flexible route search.
(2) In the route search apparatus of the above aspect, the route
searcher may calculate the overall cost value by adding a
correction value calculated as a product of the weight coefficient
and a value having a positive correlation to an integrated value of
the variance values corresponding to links that are passed through
between the place of departure and the destination, to an
integrated value of the average cost values corresponding to the
links, in a plurality of route candidates that are candidates of
the recommended route. The route search apparatus of this aspect
can readily calculate the overall cost value using a predetermined
function.
(3) In the route search apparatus of the above aspect, the route
searcher may determine a route from the place of departure to a
node corresponding to a specific point in the middle of the route
from the place of departure to the destination, as a halfway route
of the recommended route, based on a candidate overall cost value
that is provided as a sum of a first term representing an
integrated value of the average cost values corresponding to links
that are passed through from the place of departure to the node
corresponding to the specific point in the middle of the route from
the place of departure to the destination and a second term
representing a correction value calculated based on the weight
coefficient and an integrated value of the variance values
corresponding to the links that are passed through. The route
search apparatus of this aspect may determine the halfway route
based on the candidate overall cost value that is the sum of the
integrated value of the average cost values and the correction
value. This configuration can determine the halfway route by taking
into account the weight coefficient of each link.
(4) In the route search apparatus of the above aspect, the route
searcher may determine, as the halfway route, a halfway route
candidate having a smallest candidate overall cost value out of a
plurality of the candidate overall cost values, among a plurality
of halfway route candidates that are candidates of the halfway
route. The route search apparatus of this aspect may determine the
halfway route candidate having the smallest candidate overall cost
value after addition of the correction value, as the halfway route.
This configuration simplifies the process of determining the
halfway route.
(5) In the route search apparatus of the above aspect, when there
are two or more candidate overall cost values that are different
from each other by at most a predetermined value, out of the
candidate overall cost values of a plurality of halfway route
candidates that are candidates of the halfway route, the route
searcher may determine the halfway route, based on one of the first
term and the second term that is selected according to the weight
coefficient. The route search apparatus of this aspect can flexibly
determine the recommended route from the place of departure to the
destination, based on the set weight coefficient. For example, in
the case of a small weight coefficient, more emphasis is placed on
the average cost value than the variance value, and a route having
a smaller integrated value of the average cost values may be
determined as the recommended route. In the case of a large weight
coefficient, more emphasis is placed on the variance value than the
average cost value, and a route having a smaller integrated value
of the variance values may be determined as the recommended
route.
(6) In the route search apparatus of the above aspect, the route
searcher may perform a first determination process that determines
a first halfway route candidate having a smallest candidate overall
cost value out of a plurality of the candidate overall cost values,
among a plurality of halfway route candidates that are candidates
of the halfway route; a second determination process that is
performed when there is at least one provisional second halfway
route candidate having a smaller integrated value of the average
cost values than an integrated value of the average cost values of
the first halfway route candidate, out of remaining halfway route
candidates that are the halfway route candidates other than the
first halfway route candidate, and determines a second halfway
route candidate having a smallest candidate overall cost value out
of at least one provisional second halfway route candidate; and a
third determination process that specifies the second halfway route
candidate determined by the second determination process, as the
first halfway route candidate, specifies the halfway route
candidate other than the determined first halfway route candidate
and second halfway route candidate, as the remaining halfway route
candidate, and repeats the second determination process. The route
searcher may determine the first halfway route candidate and the
second halfway route candidate determined by the first to the third
determination processes, as the halfway routes. The route search
apparatus of this aspect can more accurately determine a route
having a smallest overall cost value as the recommended route.
(7) In the route search apparatus of the above aspect, when there
are a plurality of halfway route candidates that are candidates of
a halfway route from the place of departure to a node corresponding
to a specific point in the middle of the route from the place of
departure to the destination, the route searcher may process
statistical information indicating histograms of the travel time of
respective links corresponding to roads that are passed through in
each of the halfway route candidates, by convolution operation, so
as to generate candidate statistical information indicating a
histogram of the travel time with regard to each of the halfway
route candidates. The route searcher may determine the halfway
route out of the plurality of halfway route candidates, based on a
candidate overall cost value calculated according to a function
including the weight coefficient and a candidate average cost value
representing an average of the travel time of each of the halfway
route candidates and a candidate variance value representing a
degree of variance of the travel time of the halfway route
candidate that are calculated from the candidate statistical
information. The route search apparatus of this aspect may use the
candidate variance value calculated from the candidate statistical
information after the convolution operation to calculate the
candidate overall cost value. This configuration can calculate the
candidate overall cost value using a more accurate variance value
having a reduced error.
(8) In the route search apparatus of the above aspect, the route
searcher may calculate the candidate overall cost value according
to a function including a first term representing the candidate
average cost value and a second term representing a correction
value calculated based on the candidate variance value and the
weight coefficient. The route search apparatus of this aspect can
readily calculate the candidate overall cost value using the
function including the candidate average cost value and the
correction value.
(9) In the route search apparatus of the above aspect, when there
are a plurality of the candidate overall cost values that are
different from each other by at most a predetermined value, out of
the candidate overall cost values of the plurality of halfway route
candidates, the route searcher may determine the halfway route
based on one of the first term and the second term selected
according to the weight coefficient. The route search apparatus of
this aspect can flexibly determine the recommended route from the
place of departure to the destination, based on the set weight
coefficient. For example, in the case of a small weight
coefficient, more emphasis is placed on the candidate average cost
value than the candidate variance value, and a route having the
smaller candidate average cost value may be determined as the
recommended route. In the case of a large weight coefficient, more
emphasis is placed on the candidate variance value than the
candidate average cost value, and a route having the smaller
candidate variance value may be determined as the recommended
route.
(10) In the route search apparatus of the above aspect, the route
searcher may determine the recommended route with regard to each of
a plurality of different values of the weight coefficient. The
route search apparatus of this aspect may determine the recommended
route with regard to each value of the weight coefficient and can
thus inform the user of a plurality of recommended routes having
different values of the weight coefficient.
(11) In the route search apparatus of the above aspect, the route
searcher may process statistical information indicating histograms
of the travel time of respective links by convolution operation, so
as to generate statistical information indicating a histogram of
the travel time of the recommended route, and may calculate an
index indicating a degree of variance of the travel time of the
recommended route, based on a standard deviation of the generated
statistical information. The route search apparatus of this aspect
may calculate the standard deviation from the statistical
information after the convolution operation and can thus calculate
the more accurate degree of variance of the travel time with regard
to the recommended route.
(12) In the route search apparatus of the above aspect, the average
cost value and the variance value with regard to each of the links
may be calculated based on original information regarding travel
time data of the travel time and a probability of each travel time.
When the travel time of a link is affected by a feature at a
certain frequency, the average cost value of a specific link that
is the link affected by the feature may be calculated from the
entire travel time data and all the probabilities included in the
original information, and the variance value of the specific link
may be calculated from the travel time data and the probability
that are estimated to be not affected by the feature in the
original information. The route search apparatus of this aspect can
provide the average cost value that accurately reflects the travel
time data of the original information, while correcting the
variance value that is made excessive by the effect of the
feature.
The invention may be implemented by various aspects, for example, a
route search method, a route search system, a computer program or
data configured to implement any of the apparatus, the method or
the system, and a non-transitory physical recording medium in which
the computer program of data is recorded, in addition to the route
search apparatus.
BRIEF DESCRIPTION OF DRAWINGS
FIG. 1 is a diagram illustrating the configuration of a route
search system;
FIG. 2 is a diagram illustrating network data;
FIG. 3 is a diagram showing the detailed structure of link
data;
FIG. 4 is a diagram showing the detailed structure of node
data;
FIG. 5 is a flowchart showing a route search process;
FIG. 6 is a diagram showing a concrete example of the route search
process;
FIG. 7 is a diagram showing an output information window displayed
on a display panel;
FIG. 8 is a flowchart showing a route search process according to a
second embodiment;
FIG. 9 is a flowchart showing a route search process according to a
third embodiment of the invention;
FIG. 10 is a first diagram illustrating the flowchart of FIG.
9;
FIG. 11 is a second diagram illustrating the flowchart of FIG.
9;
FIG. 12 is a third diagram illustrating the flowchart of FIG.
9;
FIG. 13 is a conceptual view showing a convolution operation;
and
FIG. 14 is a diagram illustrating a method of calculating an
average cost value and a variance value.
DESCRIPTION OF EMBODIMENTS
A. First Embodiment
FIG. 1 is a diagram illustrating the configuration of a route
search system 10 according to a first embodiment of the invention.
The route search system 10 includes a route server 20 provided as a
route search apparatus and a car navigation system 50 mounted on an
automobile 12. Both the route server 20 and the car navigation
system 50 are connected to the Internet INT. The car navigation
system 50 is wirelessly connected to the Internet INT via a base
station BS. The route search system 10 is a system configured to
display a recommended route from a set place of departure to a set
destination in a visible manner on a display panel 65 included in
the car navigation system 50.
The car navigation system 50 includes a GPS receiver 69, a main
controller 51, an operating part 67, a communicator 61, an audio
output part 63 and a display panel 65. The GPS receiver 69
receives, in the form of radio wave, information for identifying
the current location (latitude and longitude) of the car navigation
system 50 measured by using satellites included in GPS (global
positioning system).
The display panel 65 includes a liquid crystal display and a drive
circuit configured to drive the liquid crystal display. The display
panel 65 is not necessarily limited to the liquid crystal display,
but any of various display devices such as organic EL display may
be employed for the display panel 65. The display panel 65 causes
the user to visually recognize various information including a
place of departure and a destination. A search setting window W1
displayed on the display panel 65 includes a field SL for entering
the place of departure, a field DL for entering the destination and
a field AI for entering additional information. The user operates
the operating part 67 to fill the respective fields SL, DL and AI.
The additional information herein denotes information regarding the
degree of accuracy in route search to be performed by the route
server 20. More specifically, this information regards a weight
coefficient .lamda. for a variance value described later. The
search setting window W1 is configured to allow for selection of
one or a plurality of options among three options "quick",
"standard" and "accurate".
The following relationships may be provided between the three
options shown in the field AI of additional information and the
weight coefficient .lamda.:
(i) option "quick". The weight coefficient .lamda. is set to "0",
and a shortest route having the shortest average travel time among
a plurality of routes from the place of departure to the
destination is determined irrespective of the variance value as a
recommended route by the route server 20;
(ii) option "standard": The weight coefficient .lamda. is set to
"1", and a route placing more emphasis on the variance value than
the option "quick" among the plurality of routes from the place of
departure to the destination is determined as a recommended route
by the route server 20; and (iii) option "accurate": The weight
coefficient .lamda. is set to "2", and a route placing more
emphasis on the variance value than the option "standard" among the
plurality of routes from the place of departure to the destination
is determined as a recommended route by the route server 20.
The weight coefficient .lamda., is not limited to the three levels
"0", "1" and "2", but may be set to a plurality of integral numbers
or may be set to continuous numerical values by using a bar or the
like.
The audio output part 63 is comprised of, for example, a speaker
configured to output voice and a drive circuit configured to drive
the speaker. The communicator 61 makes wireless data communication
or voice communication with the base station BS. The operating part
67 is an input device comprised of, for example, a numeric keypad,
arrow keys and a touch panel. The operating part 67 receives inputs
of various information for route search, for example, the place of
departure and the destination.
The main controller 51 controls the operations of the respective
components of the car navigation system 50. The main controller 51
includes a CPU 52, a RAM 54 and a ROM 56. The CPU 52 loads and
executes a program stored in the ROM 56, on the RAM 54 to implement
functions for performing various processes. For example, the main
controller 51 controls the display panel 65 to show a map image, a
recommended route and the current location. The main controller 51
also controls the communicator 61 to make communication with the
route server 20 via the base station BS. The main controller 51 may
measure current location information of the car navigation system
50 using the GPS via the GPS receiver 69 at predetermined time
intervals to generate information indicating the place of
departure.
The route server 20 is a server configured to search a route from a
place of departure to a destination specified by the car navigation
system 50 in response to a route search request from the car
navigation system 50 and send output information indicating a
search result via the Internet INT to the car navigation system 50.
In the description below, search of a route from a place of
departure to a destination performed by the route server 20 is
called route search process. The route server 20 includes a
communicator 21, a controller 22, a route database 23 (also called
route DB 23) as a memory unit (memory device) and a map database 28
(also called map DB 28). The communicator 21 makes communication
with the car navigation system 50 via the Internet INT. The
controller 22 controls the operations of the route server 20. The
route DB 23 stores road network data 24 that shows a road network
on a map by network data. The road network data 24 includes link
data 25 and node data 26. The node data 26 specifies a plurality of
nodes representing reference points on roads. The link data 25
specifies a plurality of links connecting the plurality of nodes
specified by the node data 26. The details of the link data 25 and
the node data 26 will be described later. The map DB 43 stores map
data that is to be supplied to the car navigation system 50, in a
vector data format. The map data may be stored a raster data format
such as bitmap format or JPEG format, in place of the vector
format. This map data includes data regarding the configuration of
features such as land features, buildings and roads.
FIG. 2 is a diagram illustrating road network data NW1 indicating
roads in a predetermined area as a concrete example of the road
network data 24. The road network data NW1 is data showing an
arrangement of roads by links and nodes. In the description below,
each node in the drawings may be expressed individually by a sign
including an alphabetical letter "N" with a number as the suffix,
and each link in the drawings may be expressed individually by a
sign including an alphabetical letter "L" with a number as the
suffix. FIG. 2 shows four nodes Ni to N4 and four links L1 to L4.
The nodes N1 to N4 indicate characteristic reference points on
roads or lanes. This reference point may be, for example, an
intersection, a road junction or a point where the width of the
road starts changing. The links L1 to L4 indicate roads or lanes
that interconnect the nodes N1 to N4. Route search information RSI
used to search a route from a place of departure S to a destination
point G is specified corresponding to each of the links L1 to L4.
The route search information RSI includes an average cost value AC
indicating an average of travel time with regard to each of the
links L1 to L4 and a variance value VV indicating a variance of the
travel time. The average cost value AC may be calculated from a
histogram that is data shown by the travel time as abscissa and the
probability (%) of each travel time as ordinate. According to this
embodiment, this histogram is generated based on probe data
collected from probe cars via a network. According to this
embodiment, the variance value VV denotes a dispersion calculated
from the histogram. According to another embodiment, the variance
value VV may be a standard deviation in place of the dispersion. In
the road network data NW1 of FIG. 2, a first numerical value in
parentheses with regard to each of the links L1 to L4 shows the
average cost value AC, and a second numerical value shows the
variance value VV. Traffic control information regarding traffic
regulations is also specified in the road network data NW1. The
traffic control information includes, for example, information
indicating no left-turn from the link L3 to the link L2.
FIG. 3 is a diagram showing the detailed structure of the link data
25 in the road network data 24. The link data 25 includes link
attribute data 34 showing the attribute of each link. The attribute
of a link shown by the link attribute data 34 includes a link
number, a starting point node, an end point node, an average cost
value AC and a variance value VV.
The link number of the link attribute data 34 denotes a unique
number assigned to each link for identification of the link. The
starting point node of the link attribute data 34 denotes a sign
for identifying a node with which the link is connected as the
starting point. The end point node of the link attribute data 34
denotes a sign for identifying a node with which the link is
connected as the end point. The average cost value AC of the link
attribute data 34 indicates the average of the travel time of the
link. The variance value VV of the link attribute data 34 indicates
the degree of variance of the travel time of the link. The
illustrated example of FIG. 3 shows the detailed contents of the
link attribute data 34 with regard to the link L2 to which the link
number "L2" is assigned. More specifically, the link attribute data
34 shows that the link L2 connects the "starting point node N2" to
the "end point N3", the average cost value AC of the link L2 is 16
minutes, and the variance value VV of the link L2 is 3.
FIG. 4 is a diagram showing the detailed structure of the node data
26 in the road network data 24. The node data 26 includes node
attribute data 31 showing the attribute of each node. The attribute
of a node shown by the node attribute data 31 includes a node
number, position coordinates, a node type, the number of connecting
links and connecting link numbers.
The node number of the node attribute data 31 denotes a unique
number assigned to each node for identification of the node. The
position coordinates of the node attribute data 31 indicate the
position of the node on the map. The node type of the node
attribute data 31 denotes the type of a reference point shown by
the node. The number of connecting links of the node attribute data
31 indicate the number of links connecting with the node. The
connecting link numbers of the node attribute data 31 denote
information for identifying the links connecting with the node. The
illustrated example of FIG. 4 shows the detailed contents of the
node attribute data 31 with regard to the node N2 to which the node
number "N2" is assigned. More specifically, the node attribute data
31 shows that the node N2 is located at the coordinates "Xn2
(longitude), Yn2 (latitude)), the node N2 represents an
"intersection", the number of links connecting with the node N2 is
"2", and the connecting link numbers are "L1, L2".
FIG. 5 is a flowchart showing a route search process performed by a
route searcher 29 of the route server 20. The route search process
is started when the route server 20 receives startup information of
the route search process from the car navigation system 50. The
startup information includes point information regarding a place of
departure and a destination set by the user using the car
navigation system 50, and coefficient information regarding the
weight coefficient .lamda.. More specifically, the user uses the
car navigation system 50 to enter a place of departure, a
destination and additional information regarding the weight
coefficient .lamda., and uses the operating part 67 to provide the
route server 20 with an instruction to start the route search
process. The information regarding the place of departure may not
be generated by the user's entry but may be automatically generated
based on information regarding the place of departure received by
the GPS receiver 69 of the car navigation system 50.
On the start of the route search process, the route searcher 29
sets the coordinates of the place of departure and the coordinates
of the destination used in the route search process, based on the
point information included in the startup information (step S12).
After step S12, the route searcher 29 sets a point of departure S
as the starting point of a route and a destination point G as the
end point of the route in the route search process, based on the
coordinates of the place of departure and the coordinates of the
destination (step S14) In an example described in this embodiment,
the node N1 is set as the point of departure S, and the node N4 is
set as the destination point G. When the set place of departure or
the set destination is not located at a node, a point on a link
nearest to the set place of departure or the set destination (may
be called lead-in point) is set as the point of departure S or the
destination point G. The route searcher 29 subsequently determines
a route having a smallest overall cost value that is a summation of
the route passed through as a recommended route among routes
possibly taken from the point of departure S to the destination
point G. The overall cost value denotes the sum of an integrated
value of average cost values AC corresponding to links which are
passed through from the point of departure S to the destination
point G and a correction value calculated based on the weight
coefficient .lamda.and an integrated value of variance values VV
corresponding to the links which are passed through. More
specifically, according to this embodiment, the overall cost value
is determined by Equation (1) given below:
[Math. 1] Overall cost value =.SIGMA.A +.lamda. {square root over
(.SIGMA.V)} (1)
where A denotes the average cost value AC of each of the links on a
route from the point of departure S to the destination point G;
.lamda. denotes the weight coefficient; and V denotes the variance
value VV of each of the links on the route from the point of
departure S to the destination point G.
After step S14, the route searcher 29 sets departure point
information regarding the point of departure S (step S16). The
departure point information indicates an average cost value AC and
a variance value VV from the point of departure S to a next node.
When the point of departure S is located at a node, both the
average cost value AC and the variance value VV are set to zero.
When the point of departure S is located on a link, the average
cost value AC and the variance value VV corresponding to the link
on which the point of departure S is located are calculated and set
by division using a ratio of a distance from the point of departure
S to an end point of the link to a distance from a starting point
to the end point of the link. According to this embodiment, both
the average cost value AC and the variance value VV at the point of
departure S are set to zero.
After step S16, the route searcher 29 sets the weight coefficient
.lamda. (step S17). The weight coefficient .lamda., is set, based
on the coefficient information with regard to the weight
coefficient .lamda., included in the startup information supplied
from the car navigation system 50. When a plurality of values are
set to the weight coefficient .lamda., the route searcher 29
selects an arbitrary value of the weight coefficient .lamda. and
performs subsequent steps. After step S17, the route searcher 29
generates a candidate label that is an index for determining links
which are to be passed through on a route from the point of
departure S to the destination point G (step S18). When the
candidate label is generated for a certain link located in the
middle of the route from the point of departure S to the
destination point G, the candidate label is set at an end point
(node) of the certain link. The candidate label is comprised of an
integrated value of the average cost values AC of the respective
links between the point of departure S and the certain link and an
integrated value of the variance values VV of the respective links
between the point of departure S and the certain link. A candidate
overall cost value is then calculated, based on the information
included in the candidate label. More specifically, the candidate
overall cost value is calculated according to Equation (2) given
below:
[Math. 2] Candidate overall cost value=.SIGMA.A1+.lamda. {square
root over (.SIGMA.V1)} (2) where A1 denotes the average cost value
AC of each of the links on a route from the point of departure S to
a predetermined node that is an end point in the middle of the
route; .lamda. denotes the weight coefficient; and V1 denotes the
variance value VV of each of the links which are passed through
from the point of departure S to the predetermined node that is the
end point in the middle of the route.
After step S18, the route searcher 29 determines a candidate label
having a smallest candidate overall cost value among at least one
candidate overall cost value, as a fixed label (step S20).
Determining the fixed label fixes a route (halfway route) to a node
(temporary fixed node) that is located in the middle of the route
from the point of departure S to the destination point G. The route
searcher 29 subsequently determines whether a last link or a node
(last node) that is an end point of the last link in the halfway
route toward the destination point G is a link or a node where the
destination point G is located (step S22). When it is determined
that the last link or the last node is the link or the node where
the destination point G is located, the route searcher 29 fixes the
halfway route as a recommended route. The route searcher 29 then
generates output information to display the fixed recommended route
on the display panel 65 of the car navigation system 50 (step S23).
More specifically, the output information includes information
regarding the recommended route from the point of departure S to
the destination point G, information regarding an average travel
time from the point of departure S to the destination point G and
variance information regarding a variance of the average travel
time. The details of this output information will be described
later. After fixing the recommended route, the route searcher 29
determines whether the recommended route has been fixed with regard
to all the values of the weight coefficient .lamda. included in the
startup information (step S24). When it is determined that the
recommended route has been fixed with regard to all the values of
the weight coefficient .lamda., the route searcher 29 terminates
the route search process.
When it is determined that the last link or the last node is not
the link or the node where the destination point G is located, the
route searcher 29 further extends the search tree from the end
point of the halfway route toward the destination point G by the
Dijkstra's algorithm and generates a candidate label (step S18).
The route searcher 29 then performs the series of processes of and
after step S20 again. When it is determined that the recommended
route has not yet been fixed with regard to all the values of the
weight coefficient .lamda., the route searcher 29 sets another
value of the weight coefficient .lamda. for which the recommended
route has not yet been fixed at step S17 and performs the
subsequent series of processes again.
FIG. 6 is a diagram showing a concrete example of the route search
process. Steps shown in FIG. 6 correspond to the steps shown in
FIG. 5. Sub-steps shown in FIG. 6 show concrete processes performed
at the respective steps shown in FIG. 5. FIG. 6 illustrates a
concrete example of the route search process when the route
searcher 29 sets the node N1 shown in FIG. 2 as the point of
departure S and sets the node N4 shown in FIG. 2 as the destination
point G. The weight coefficient .lamda., is set to "1" at step S17.
Sub-steps C1 and C2 respectively correspond to steps S14 and
S16.
The route searcher 29 extends the search tree from the point of
departure S toward the destination point G by the Dijkstra's
algorithm. More specifically, the route searcher 29 (shown in FIG.
1) refers to the node data 26 and the link data 25 and sets a
candidate label from the point of departure S to an end point of a
next link as shown in FIG. 6 (step S18). In the concrete example,
first links from the node N1 as the point of departure S toward the
destination point are the links L1 and the links L3. The route
searcher 29 sets a candidate label T1 for a route from the point of
departure S to an end point of the link L1 (or more specifically,
at the end point of the link L1). The route searcher 29 then
calculates a candidate overall cost value V1 based on the candidate
label T1. The route searcher 29 also sets a candidate label T2 for
a route from the point of departure S to an end point of the link
L3 (sub step C4). The route searcher 29 then calculates a candidate
overall cost value V2 based on the candidate label T2.
At sub-step C3, a candidate overall cost value of the route from
the point of departure S to a next node N2 is calculated according
to Equation (2) given above. More specifically, an integrated value
of average cost values AC (integrated cost value) "15" is
calculated by summing up an average cost value AC "0" set at the
point of departure S and an average cost value AC "15" set at the
link L1. At sub-step C3, an integrated value of variance values VV
(integrated variance value) "3" is also calculated by summing up a
variance value VV "0" set at the point of departure S and a
variance value VV "3" set at the link L1. The route searcher 29
subsequently adds a correction value that is the product of the
positive square root of the integrated variance value "3" and the
weight coefficient .lamda., to the integrated cost value "15", so
as to calculate a candidate overall cost value V1. The calculated
candidate overall cost value V1 is equal to "16.4". According to
this embodiment, the candidate overall cost value is rounded off to
one decimal place.
At sub-step C4, a candidate overall cost value of the route from
the point of departure S to a next node N3 is calculated according
to Equation (2) given above. More specifically, an integrated cost
value "30" is calculated by summing up the average cost value AC
"0" set at the point of departure S and an average cost value AC
"30" set at the link L3. At sub-step C4, an integrated variance
value VV "15" is also calculated by summing up the variance value
VV "0" set at the point of departure S and a variance value "15"
set at the link L3. Like sub-step C3, the route searcher 29
subsequently adds a correction value that is the product of the
positive square root of the integrated variance value "15" and the
weight coefficient .lamda. to the integrated cost value "30", so as
to calculate a candidate overall cost value V2. The calculated
candidate overall cost value V2 is equal to "37.7".
After calculation of the candidate overall cost values V1 and V2
from the point of departure S to the next nodes, a candidate label
having the smaller candidate overall cost value between the
candidate overall cost values V1 and V2 is determined as a fixed
label. At sub-step C5, the candidate label having the candidate
overall cost value V1 is determined as a fixed label. Accordingly
the route from the point of departure S to the link L1 is fixed as
a halfway route. When it is subsequently determined that the link
L1 or the end point of the link L1 (node N2) is not the link or the
node where the destination point G is located, the route searcher
29 further extends the search tree by the Dijkstra's algorithm to
generate a candidate label (at sub-step C6). More specifically, the
route searcher 29 refers to the node data 26 and the link data 25
to specify the link L2 from the node N2 toward the destination
point G, and generates a candidate label T3 for a route from the
point of departure S through the link L1 and the node N2 to the
link L2. A candidate overall cost value V3 of the candidate label
T3 is then calculated according to Equation (2) given above, like
sub-steps C3 and C4. The candidate overall cost value V3 of the
candidate label T3 generated at sub-step C6 is equal to "35.9". The
candidate label T3 having the candidate overall cost value V3 is
set at an end point of the link L2 (node N3). There are two
candidate labels T2 and T3 set at the node N3 by the processes of
sub-step C4 and sub-step C6. The route searcher 29 then compares
the candidate overall cost values V2 and V3 of the two candidate
labels T2 and T3 and determines the candidate label T3 having the
smaller candidate overall cost value V3 as a fixed label, while
deleting the other candidate label T2. Accordingly the route from
the point of departure S through the link L1, the node N2 and the
link L2 to the node N3 is fixed as a halfway route. When it is
subsequently determined that the link L2 or the end point of the
link L2 (node N3) is not the link or the node where the destination
point G is located, the route searcher 29 further extends the
search tree by the Dijkstra's algorithm to generate a candidate
label T4 (at sub-step C8). More specifically, the route searcher 29
refers to the node data 26 and the link data 25 to specify the link
L4 from the node N3 toward the destination point G, and generates a
candidate label T4 for a route from the point of departure S
through the link L1, the node N2, the link L2 and the node N3 to
the link L4. There is only one label T4 generated by extending the
search tree from the fixed label determined at previous sub-step
C7, and there is no other candidate label. The route searcher 29
then determines the candidate label T4 as a fixed label.
Accordingly the route from the point of departure S to the link L4
having the candidate label T4 determined as the fixed label is
fixed as a halfway route. Since the destination point G is the node
N4, the destination point G is located at the link L4 or the end
point of the link L4 (node N4) in the fixed halfway route. The
route searcher 29 accordingly determines the halfway route fixed at
sub-step C8 as a recommended route from the point of departure S to
the destination point G (at sub-step C9). At sub-step C9, the route
searcher 29 also refers to the link data 25 to integrate the
average cost values AC corresponding to the links L1, L2 and L4 of
the recommended route and thereby calculates an average travel time
of the recommended route. According to this embodiment, the average
travel time is "46 minutes". The route searcher 29 also refers to
the link data to calculate an integrated value of variance values
(integrated variance value) corresponding to the links L1, L2 and
L4 of the recommended route. The positive square root of the
integrated variance value is provided as a variance index (standard
deviation) indicating the degree of variance of the average travel
time. According to this embodiment, the variance index is "4".
FIG. 7 is a diagram illustrating an output information window W2
displayed on the display panel 65. According to this embodiment, it
is assumed that the user selects two options "standard (.lamda.=1)"
and "quick (.lamda.=0)" in the search setting window "1. The output
information window W2 includes recommended routes and values of a
travel time determined for respective values of the weight
coefficient .lamda.. The recommended route is shown by providing
marks such as red lines on map data. The travel time is shown by
the average travel time and the standard deviation. More
specifically, with regard to "route (standard)" having the weight
coefficient .lamda.=1, the travel time is shown by a formula
(46.+-.4) that is the average travel time of 46 minutes plus minus
the standard deviation "4". When the user selects a recommended
route between the displayed two routes (standard, quick), the car
navigation system 50 starts a route guidance.
As described above, the first embodiment determines the recommended
route from the point of departure S representing the place of
departure to the destination point G representing the destination,
based on the average cost value AC, the variance value VV and the
weight coefficient .lamda.. This configuration allows for flexible
route search by simply changing the value of the weight coefficient
.lamda..
B. Second Embodiment
FIG. 8 is a flowchart showing a route search process according to a
second embodiment of the invention. The route search process of the
second embodiment differs from the route search process of the
first embodiment (shown in FIG. 5) by the details of the process of
determining a candidate label as a fixed label. FIG. 8 accordingly
shows the details of a process of determining a fixed label (step
S20a) in the route search process. The other processes and the
configuration of the route search system 10 are similar to those of
the first embodiment and are not specifically described here. At
step S20 of the first embodiment, a label having the smallest value
among the candidate overall cost values is determined as a fixed
label. At step S20a of the second embodiment, on the other hand, a
fixed label is determined based on the set weight coefficient
.lamda., when a predetermined condition is satisfied as described
below in detail.
At step S20a, the route searcher 29 first compares the candidate
overall cost values of the generated candidate labels (step S40).
When there are a plurality of candidate overall cost values, the
route searcher 29 determines whether a difference between the
candidate overall cost values is equal to or less than a
predetermined value (step S42). According to this embodiment, the
route searcher 29 extracts two candidate overall cost values or
more specifically the smallest and the second smallest candidate
overall cost values among the plurality of candidate overall cost
values, and calculates a difference between the two extracted
candidate overall cost values. According to this embodiment, the
predetermined value is set to "0.2". The predetermined value may,
however, be equal to "0" or may be equal to a numerical value other
than 0.2. When it is determined that the difference between the
candidate overall cost values is greater than the predetermined
value, a candidate label having the smallest candidate overall cost
value is determined as a fixed label (step S44). When it is
determined that the difference between the candidate overall cost
value is equal to or less than the predetermined value, on the
other hand, a fixed label is determined, based on the weight
coefficient .lamda., (step S46). More specifically, in a first case
having the small weight coefficient .lamda., a candidate label
having the minimum integrated cost value (first term on the right
side of Equation (2)) among the plurality of candidate labels is
determined as a fixed label. In a second case having the larger
weight coefficient .lamda. than that in the first case, a candidate
label having the minimum integrated variance value (second term on
the right side of Equation (2)) is determined as a fixed label.
According to this embodiment, the case having the weight
coefficient .lamda. equal to "0" or "1" corresponds to the first
case, and the case having the weight coefficient .lamda. equal to
"2" corresponds to the second case.
As described above, the second embodiment flexibly determines the
recommended route from the point of departure S to the destination
point G, based on the set weight coefficient .lamda.. For example,
in the first case having the small weight coefficient .lamda., more
emphasis is placed on the average cost value AC than the variance
value VV, and the route having the smallest integrated value of the
average cost values AC is determined as the recommended route. In
the second case having the larger weight coefficient .lamda. than
that in the first case, more emphasis is placed on the variance
value VV than the average cost value AC, and the route having the
smallest integrated value of the variance values is determined as
the recommended route.
C. Third Embodiment
FIG. 9 is a flowchart showing a route search process according to a
third embodiment of the invention. FIG. 10 is a first diagram
illustrating the flowchart of FIG. 9. FIG. 11 is a second diagram
illustrating the flowchart of FIG. 9. FIG. 12 is a third diagram
illustrating the flowchart of FIG. 9. FIGS. 10 and 12 illustrate
road network data NW2 used for the purpose of describing the third
embodiment. According to the third embodiment, a node N6 is set as
the point of departure S, a node N30 is set as the destination
point G, and there are routes passing through links L5 to L14 in
the middle from the point of departure S to the destination point
G. In the network data NW2, nodes N6 to N11 which are
interconnected by the links L5 to L14 are also illustrated.
The route search process of the third embodiment differs from the
route search process of the first embodiment (shown in FIG. 5) by
the details of the process of determining a candidate label as a
fixed label and a process of determining a recommended route when
reaching the destination point G. FIG. 9 accordingly shows the
details of a process of determining a fixed label (step S20b) in
the route search process. Among the other processes of the third
embodiment, the processes similar to those of the first embodiment
are not specifically described here. The configuration of the route
search system 10 (shown in FIG. 1) is similar to that of the first
embodiment. At step S20b of the third embodiment, in addition to a
first candidate label having the smallest candidate overall cost
value among a plurality of candidate labels, a second candidate
label that satisfies a predetermined condition is determined as a
fixed label. In the description below, it is assumed that the
weight coefficient .lamda. is set to "2". As shown in the field of
the processing details of step S52 in FIG. 10, candidate labels T5
to T8 respectively corresponding to routes R5 to R8 from the point
of departure S to a node N10 are set at the node N10, in order to
determine a halfway route from the point of departure S to the node
N10. An integrated value of average cost values AC (integrated cost
value), an integrated value of variance values VV (integrated
variance value), and a candidate overall cost value calculated
according to Equation (2) given above with regard to each of the
candidate labels T5 to T8 are also shown in the field of the
processing details of step S52 in FIG. 10.
As shown in FIG. 9, at step S20b, the route searcher 29 first
compares the candidate overall cost values of the generated
candidate labels T5 to T8 (shown in FIG. 10) (step S52), and
subsequently determines a candidate label having the smallest
candidate overall cost value among the plurality of candidate
overall cost values, as a first candidate label (step S54). In the
illustrated example of FIG. 10, the candidate label T5 having the
candidate overall cost value of "35.0" is determined as the first
candidate label, so that the route R5 corresponding to the
candidate label T5 is determined as a first halfway route candidate
R5 (step S54).
As shown in FIG. 9, the route searcher 29 subsequently determines
whether there is any provisional second halfway route candidate
(step S56). The provisional second halfway route candidate denotes
a route candidate having a smaller integrated cost value than the
integrated cost value of the first halfway route candidate R5 among
the halfway route candidates R6 to R8 other than the first halfway
route candidate R5 (remaining halfway route candidates R6 to R8) as
shown in FIG. 10. In the illustrated example of FIG. 10, all the
integrated cost values "18, 19, 25" of the candidate labels T6 to
T8 are smaller than the integrated cost value "29" of the candidate
label T5, so that the halfway route candidates R6 to R8
corresponding to the candidate labels T6 to T8 are the provisional
second halfway route candidates R6 to R8.
The route searcher 29 then compares the candidate labels T6 to T8
corresponding to the provisional second halfway route candidates R6
to R8 (step S58) and determines a candidate label having the
smallest candidate overall cost value as a second candidate label
(step S60). In the illustrated example of FIG. 10, the candidate
label T7 having the candidate overall cost value of "47.3" is
determined as the second candidate label, so that the route R7
corresponding to the candidate label T7 is determined as a second
halfway route candidate R7 (step S60).
The route searcher 29 subsequently specifies the halfway route
candidates R6 and R8 other than the previously determined first and
second halfway route candidates R5 and R7 as remaining halfway
route candidates R6 and R8 and performs the processing of step S56.
When the first and the second halfway route candidates R5 and R7
have already been determined, the route searcher 29 specifies the
second halfway route candidate R7 determined immediately before the
processing of step S56 as the first halfway route candidate R7 and
determine whether there is any route candidate having a smaller
integrated cost value than the integrated cost value "19" of the
first halfway route candidate R7. In the illustrated example of
FIG. 11, the integrated cost value "18" of the candidate label T6
is smaller than the integrated cost value "19" of the candidate
label T7, so that the halfway route candidate R6 corresponding to
the candidate label T6 is the provisional second halfway route
candidate R6.
The route searcher 29 then performs the processing of steps S58 and
S60 shown in FIG. 9. As shown in FIG. 11, there is only one
provisional second halfway route candidate R6, so that the
provisional second halfway route candidate R6 is determined as a
second halfway route candidate R6 (steps S58 and S60).
As shown in FIG. 9, after step S60, the route searcher 29 specifies
the halfway route candidate R8 other than the previously determined
first and second halfway route candidates R5 to R7 as a remaining
halfway route candidate R8 and performs the processing of step S56
again. As shown in FIG. 11, the integrated cost value "25" of the
candidate label T8 corresponding to the remaining halfway route
candidate R8 is larger than the integrated cost value "18" of the
candidate label T6 corresponding to the most recently determined
second halfway route candidate R6. As shown in FIG. 9, the route
searcher 29 then provides a negative answer "NO" at step S56 and
determines the previously determined candidate labels T5 to T7 as
fixed labels. Accordingly the first and second halfway route
candidates R5 to R7 corresponding to the fixed labels T5 to T7 are
determined as halfway routes R5 to R7 (step S62).
As shown in FIG. 12, the route searcher 29 extends the search tree
from the halfway routes R5 to R7 corresponding to the fixed labels
T5 to T7 toward the destination point G and generates candidate
labels T9 to T11. The route searcher 29 performs the processing of
steps S52 to S62 with regard to the candidate labels T9 to T11. In
the illustrated example of FIG. 12, the processing of steps S52 and
S54 is performed to select the candidate label T11 having the
smallest candidate overall cost value and determine the route R11
corresponding to the candidate label T11 as a first halfway route
candidate R11. The processing of step S56 is subsequently
performed. In the illustrated example of FIG. 12, there is the
route candidate R10 (provisional second halfway route candidate
R10) having a smaller integrated cost value than the integrated
cost value "79" of the first halfway route candidate R11 out of
remaining halfway route candidates R9 and R10, so that an
affirmative answer "YES" is provided at step S56. The processing of
steps S58 and S60 is then performed. There is only one provisional
second halfway route candidate R10, so that the provisional second
halfway route candidate R10 is determined as a second halfway route
candidate R10.
After step S60, the route searcher 29 specifies the halfway route
candidate R9 other than the previously determined first and second
halfway route candidates R11 and R10 as a remaining halfway route
candidate R9 and performs the processing of step S56 again. As
shown in FIG. 12, the integrated cost value "89" of the candidate
label T9 corresponding to the remaining halfway route candidate R9
is larger than the integrated cost value "78" of the candidate
label T10 corresponding to the most recently determined second
halfway route candidate R10. As shown in FIG. 9, the route searcher
29 then provides a negative answer "NO" at step S56 and determines
the previously determined candidate labels T11 and T10 as fixed
labels. Accordingly the first and second halfway route candidates
R11 and R10 corresponding to the fixed labels T11 and T10 are
determined as halfway routes R11 and R10 (step S62).
After determining the halfway routes, the route searcher 29
determines whether a last link or a node (last node) that is an end
point of the last link in the halfway route toward the destination
point G is a link or a node where the destination point G is
located (step S22 in FIG. 5). When it is determined that the last
link or the last node is the link or the node where the destination
point G is located, the route searcher 29 fixes the halfway route
as a recommended route. When there are a plurality of halfway
routes (for example, the halfway routes R10 and R11 at step S62 in
FIG. 12), the route searcher 29 fixes a halfway route having the
smallest candidate overall cost value among the plurality of
halfway routes, as a recommended route. For example, when the node
N11 is set as the destination point G in the network data NW2 shown
in FIG. 12, the halfway route R10 having the smaller candidate
overall cost value out of the halfway route R10 and R11 determined
at step S62 is fixed as a recommended route.
At the point of the node N10, the candidate label T5 is the first
candidate label having the smallest candidate overall cost value as
shown in FIG. 10. After the search tree is extended, however, the
candidate label T11 is the first candidate label as shown in FIG.
12. The candidate label T11 is a label by extending the search tree
from the candidate label T7 in FIG. 10. In the case of extending
the search tree with determining only a label having the smallest
candidate overall cost value as a fixed label, a route having the
smallest candidate overall cost value is likely to be not
determinable as a recommended route. As described above, however,
this embodiment determines a candidate label having the smallest
candidate overall cost value (first candidate label) and
additionally a candidate label that satisfies a predetermined
condition (second candidate label) among candidate labels having
smaller integrated cost values than the first candidate label, as
fixed labels. This configuration enables a route having the
smallest overall cost value to be determined more accurately as a
recommended route.
There is the following correspondence relationship between the
respective steps of the third embodiment and the processes
described in Summary: Steps S52 and S54 correspond to the "first
determination process"; Steps S56 to S60 correspond to the "second
determination process"; and Steps S56 to S60 performed after step
S60 correspond to the "third determination process". D.
Modifications
D-1. First Modification
In the first and the second embodiments described above, the
positive square root of the integrated variance value is calculated
as the variance index. According to a modification, statistical
information indicating histograms used for calculating the average
cost values AC of the respective links constituting a recommended
route or a halfway route may be used. More specifically,
statistical information of the respective links may be processed by
convolution operation, and a standard deviation calculated from
statistical information indicating a histogram after the
convolution operation may be used as the variance index. An average
cost value calculated from the statistical information indicating
the histogram after the convolution operation may be used to
determine a recommended route or a halfway route. The details are
described below. FIG. 13 is a conceptual view showing the
convolution operation. In the illustrated example of FIG. 13, a
recommended route from a point of departure S to a destination
point G is comprised of links L10, L12 and L14. The link data 25
includes data (statistical information) indicating histograms H10,
H12 and H14, in addition to link numbers, starting point nodes, end
point nodes, average cost values AC and variance values VV. The
histograms H10, H12 and H14 may be generated, for example, based on
probe data collected from probe cars. In the histograms H10, H12
and H14, the travel time (minutes) of a link is shown as abscissa
and the probability (%) of each travel time is shown as ordinate.
The histograms of the first link L10 and the next link L12 in a
route from the point of departure S toward the destination point G
are processed by convolution operation, and a new histogram H18 is
generated. The histogram H14 of the next link L14 subsequent to the
link L12 and the histogram H18 are processed by convolution
operation, and a new histogram H20 is generated. A standard
deviation is calculated from statistical information indicating the
histograms H18 and H20 generated by the convolution operation. In
other words, the convolution operation of the histograms is defined
by Equation (3) given below:
.times..function..times..function..times..function.
##EQU00001##
where F(m) on the left side denotes a function generated by
convolution operation of two histograms; f denotes a function
defined by the first histogram; g denotes a function defined by the
second histogram; n denotes the travel time in the first histogram;
and m denotes the travel time (total time) in convolution operation
of the first histogram and the second histogram. Like the first
embodiment and the second embodiment described above, this also
calculates the standard deviation as the variance index of the
travel time in the recommended route or the halfway route.
Calculating the standard deviation from the statistical information
indicating a histogram after the convolution operation provides a
more accurate variance index having a reduced error, compared with
the above embodiments.
The following describes a process of determining a halfway route
using the statistical information indicating the histogram after
the convolution operation according to the first modification. The
route searcher 29 (shown in FIG. 1) processes statistical
information indicating respective histograms of the travel time of
respective links that are passed through in halfway route
candidates as candidates of a halfway route by convolution
operation, so as to generate candidate statistical information
indicating histograms of the travel time in the respective halfway
route candidates. The route searcher 29 subsequently calculates a
candidate overall cost value of each halfway route candidate
according to a function including the weight coefficient .lamda.
and a candidate average cost value At1 indicating an average of the
travel time of the halfway route candidate and a candidate variance
value VV (dispersion in this modification) indicating the degree of
variance of the halfway route candidate that are calculated from
the candidate statistical information. For example, the route
searcher 29 may calculate the candidate overall cost value
according to Equation (4) given below:
[Math. 4] Candidate overall cost value=At1+.lamda. {square root
over (Vt1)} (4) where At1 denotes a candidate average cost value
calculated from the candidate statistical information; .lamda.,
denotes the weight coefficient; and Vt1 denotes a candidate
variance value VV (dispersion) calculated from the candidate
statistical information.
The right side of Equation (4) given above is defined by a first
term representing the candidate average cost value and a second
term representing a correction value as the product of the positive
square root of the candidate variance value VV and the weight
coefficient .lamda., but this is not restrictive. For example, the
second term may be the product of the candidate variance value VV
and the weight coefficient .lamda.. The right side of Equation (4)
given above may additionally include a third term and a fourth
term. For example, the third term may be defined as a term for
increasing the cost value as traffic congestion information in the
case of traffic congestion on a specific link. This process uses
the candidate variance value VV calculated from the candidate
statistical information after the convolution operation to
calculate the candidate overall cost value. This enables the
halfway route to be determined using the candidate overall cost
value calculated from the more accurate variance value VV having a
reduced error.
D-2. Second Modification
FIG. 14 is a diagram showing a method of calculating an average
cost value AC and a variance value VV with regard to a specific
link L. A graph 14A in FIG. 14 is generated based on probe data and
has the travel time (minutes) of the link L as abscissa and the
number of samples n (probability) of each travel time as ordinate.
The probability is calculated based on the number of samples. A
graph 14B is a conceptual diagram of a normal distribution based on
the average value of the information expressed by the graph 14A
(original information) and the variance value determined from
information after deletion of data of the travel time estimated to
be affected by a feature from the graph 14A. A graph 14C is a
conceptual diagram of a normal distribution based on the
information expressed by the graph 14A (original information).
When the link L (specific link L) has a specific feature such as a
traffic light or a railway crossing that affects the travel time or
when a link adjacent to the specific link L has a specific feature
such as a traffic light or a railway crossing, the specific feature
provides an effect of increasing the travel time in the specific
link L at a certain frequency. It is, however, unlikely that the
travel time is increased by the specific feature in all the links
that are passed through from a place of departure to a destination.
The average cost value AC and the variance value VV of the specific
link L may thus be calculated as described below. The average cost
value AC may be calculated from the entire travel time data
expressed by the graph 14A as the original information and their
probabilities. The variance value VV may be calculated from the
travel time data estimated to be not affected by the specific
feature and their probabilities out of the entire data expressed by
the graph 14A as the original information. In other words, the
variance value VV may be calculated based on the variance value
determined from data after deletion of data, such as the travel
time, estimated to be affected by the specific feature from the
data of the original information. For example, the "data estimated
to be affected by the specific feature" may be data of the longer
travel time than a minimum travel time at which the number of
samples becomes equal to or lower than a predetermined rate (for
example, equal to or lower than 10%) of the number of samples na
corresponding to the average cost value AC, out of the data of the
longer travel time than the average cost value AC. In another
example, the "data estimated to be affected by the specific
feature" may be data of the long travel time at a predetermined
rate or higher among the number of samples n (for example, data of
the top 10%). In FIG. 14, data that is equal to or less than the
number of samples nb out of the data having the longer travel time
than the average cost value AC is estimated as data affected by the
feature and is omitted from calculation of the variance value VV.
The normal distribution based on all the data of the original
information (graph 14C) provides the excessive variance value VV.
The normal distribution based on the information after deletion of
the data estimated to be affected by the feature (graph 14B) is, on
the other hand, appropriately corrected without providing the
excessive variance value VV. As described above, the second
modification provides the average cost value AC that accurately
reflects the travel time data of the original information, while
correcting the variance value that is made excessive by the effect
of the feature.
D-3. Third Modification
According to the first and the second embodiments described above,
the route server 20 performs the route search process, and the car
navigation system 50 receives the result of the route search
process and displays the output information on the display panel
65. This configuration is, however, not restrictive, but the output
information may be displayed on the display panel 65 by any of
various other configurations. For example, the route server 20 may
send network data of a required range including a place of
departure and a destination to the car navigation system 50. The
car navigation system 50 may receive the network data, perform the
route search process and display the output information on the
display panel 65. The car navigation system 50 may provide the user
with audio output information. The car navigation system 50 itself
may be provided with the functions of the route server 20.
D-4. Fourth Modification
The car navigation system 50 in the first and the second
embodiments described above may be replaced by any of various other
devices having the function of providing the user with output
information, for example, a cell phone or a personal computer.
D-5. Fifth Modification
In the first and the second embodiment described above, the overall
cost value and the candidate overcall cost value are calculated
according to the relational expressions of Equations (1) and (2)
given above. In Equations (1) and (2), the second term on the right
side is the product of the weight coefficient .lamda. and a value
having the positive correlation to the integrated value of the
variance value VV (more specifically, the positive square root).
These Equations (1) and (2) are, however, not restrictive, but the
overall cost value and the candidate overall cost value may be
calculated using a function including the average cost value AC,
the variance value VV and the weight coefficient .lamda.. For
example, the second term on the right side in Equation (1) given
above may be replaced by the product of the integrated value of the
variance value and the weight coefficient. The right side of
Equation (1) or (2) given above may additionally include a third
term and a fourth term. For example, the third term may be defined
as a term for increasing the cost value as traffic congestion
information in the case of traffic congestion on a specific
link.
D-6. Sixth Modification
Part of the functions implemented by the software configuration in
the above first or second embodiment may be implemented by a
hardware configuration, and part of the functions implemented by
the hardware configuration may be implemented by a software
configuration.
The invention is not limited to any of the embodiments and
modifications described above but may be implemented by a diversity
of other configurations without departing from the scope of the
invention. For example, the technical features of any of the
embodiments and modifications corresponding to the technical
features of each of the aspects described in Summary may be
replaced or combined appropriately, in order to solve part or all
of the problems described above. Any of the technical features may
be omitted appropriately unless the technical feature is described
as essential herein.
REFERENCE SIGNS LIST
10 route search system
12 automobile
18 histogram
20 route server
21 communicator
22 controller
23 route database
24 network data
25 link data
26 node data
28 map database
29 route searcher
31 node attribute data
34 link attribute data
50 car navigation system
51 main controller
52 CPU
61 communicator
63 audio output part
65 display panel
67 operating part
W1 search setting window
N1 - N4 nodes
L1 - L4, L10 - L14 links
W2 output information window
* * * * *